第五代行動通訊 (5G New Radio，5G NR) 快速發展，滿足了各種商業需求並為大量用戶提供服務。隨著通信需求之增長，毫米波頻段 (mini-meter Wave，mmWave) 被提出使用，其特性為子載波間隔 (Subcarrier Spacing，SCS) 被放大，因而導致資源排程之時隙 (Slot) 時間被縮短。此變化對於基地台 (Next Generation Node B，gNB) 內之排程器設計構成挑戰。 本研究提出了一種基於用戶設備 (User Equipment，UE) 之分散式排程器(Distributed Scheduler) 架構用於上行傳輸。透過將部分排程工作分配給UE，減輕gNB之排程負擔。gNB將排程器所需之排程資訊封裝於所設計之下行排程資訊 (Downlink Scheduling Information，DSI)，其包含UE之5G服務品質指標 (5G QoS Identifier，5QI)、緩衝狀態報告 (Buffer Status Report，BSR) 等基本控制信息，並傳送給相關 UE作為排程器所需之輸入資訊。相關UE執行排程後將產生如同gNB排程器輸出之UE所需下行控制訊息 (Downlink Control Information，DCI)。UE分別取回與自己相關之DCI，如同gNB執行排程後將對應DCI傳送給UE。因gNB未參與實際排程，因此UE傳送上行資料之頻域與時域資源之資訊需先以上行排程信息 (Uplink Scheduling Information，USI) 回傳至gNB，以完成gNB正確接收UE之傳輸資料。 為了驗證分散式排程器於5G NR之可行性，本研究使用開源軟體OpenAirInterface5G (OAI) 進行實際驗證。透過將排程器由gNB端移植至UE端，實現基於UE之分散式排程器。 ;The 5G New Radio (NR) network is developing to accommodate various business requirements and offer extensive coverage to a substantial user population. To handle the growth of traffic, the mini-meter Wave (mmWave) band is utilized, in which the Subcarrier Spacing (SCS) is increased due to the characteristic of high-frequency band. As consequence, the time slot of radio resource scheduling is decreased. The change becomes a new challenge to the scheduling design in the next generation Node B (gNB)。 This thesis proposes that all User Equipment (UE) schedules their uplink transmission in coordinated and distributed manners in order to reduce the scheduling loading in the gNB. To realize it, the essential scheduling information of the scheduler, such as 5G QoS Identifier (5QI), Buffer Status Report (BSR) and other transmission control information, is encapsulated into the Downlink Scheduling Information (DSI). The DSI is sent to the corresponding UEs for performing distributed scheduling by restoring the essential scheduling information required for the scheduler. Every UE preforms scheduling will obtain the same scheduling result, i.e., the Downlink Control Information (DCI) of corresponding UEs, as that derived by the gNB. A UE retrieving the corresponding DCI from the scheduling result looks like it receives the DCI sent from the gNB. As the gNB does not involve the scheduling, the information about the UL resource used by a UE is needed to be encapsulated into an Uplink Scheduling Information (USI) and sent from UE to gNB before UL transmission, in order to make sure that gNB is able to receive the UL data transmitted from UE. Finally, the open-source program OpenAirInterface5G (OAI) is used in this study for validating the feasibility of the proposed distributed scheduler in 5G NR. By moving the scheduler from the gNB to the UE, the UE based distributed scheduler is put into practice.
